491 results about "Neural network architecture" patented technology

The invention discloses a cable life dynamic evaluation system based on multi-physics field coupling, and particularly relates to the field of industrial automation and control systems, which comprises a multi-physics field sensing module, a coupling analysis engine module, a dynamic life evaluation module, a digital twin interaction module and an environmental interference suppression module, through a distributed optical fiber temperature sensor, a capacitive electric field sensor and a magnetostrictive stress sensor, temperature, electric field, magnetic field and mechanical stress data of a cable are collected in real time, multi-physical field characteristics and a cable defect database are matched in real time by using a cross-scale dynamic association algorithm, a damage state is evaluated, and the cable defect detection accuracy is improved. A time sequence neural network architecture is adopted to predict the remaining life, model self-correction is achieved through digital twin comparison, interference is suppressed in combination with an environment-physical field coupling compensation matrix, sensing and evaluation of the health state of the cable, life prediction and continuous optimization of the model are achieved, and the efficiency of cable life evaluation is improved.

Neural network architectures for feature extraction from visual input. In at least one embodiment, a neural network architecture for a vision backbone includes hybrid stages with at least one state space model (SSM)-based block preceding at least one transformer block. In at least one embodiment, an SSM-based block includes parallel branches, one including an SSM and one without an SSM, and a concatenation layer for concatenating the output of each branch. In at least one embodiment, the SSM performs a parallel selective scan operation to efficiently map tokens of an input sequence to tokens of an output sequence via GPU acceleration.

The invention, which relates to the technical field of hydropower station dam safety monitoring, discloses a hydropower station dam safety monitoring data acquisition and transmission system comprising a cloud twin brain module and edge neurons. The cloud twin brain module comprises a sequence neural network engine and a reflection kernel generation module, the sequence neural network engine adopts a neural network architecture with parallel and cyclic dual representation, comprises a time mixing module and a channel mixing module, and can learn a normal operation mode of the dam from historical monitoring data; and the reflection nuclear generation module compresses the reference twin model into a lightweight reflection nuclear model and issues the lightweight reflection nuclear model to the edge device. The edge neuron comprises a micro-twinborn prediction module and a hierarchical transmission control module, and the micro-twinborn prediction module predicts a theoretical expected value of a dam state in real time and calculates a reflection deviation with an actual observation value; the hierarchical transmission control module implements a three-level response strategy according to the magnitude of the reflection deviation, transmits abstract information according to an abnormal trend, and uploads an emergency abnormality in time.

The invention designs an assembly process error modeling method considering heat, and realizes rapid and accurate calculation of thermal deformation of an assembly junction surface affected by heat in the part assembly process. In the part assembling process, heat generated in the assembling process is an important influencing factor influencing the assembling precision. In order to realize rapid calculation of thermal deformation in a part assembling process, a prediction model of thermal deformation of an assembling joint surface in the part assembling process is constructed by utilizing a physical information neural network. According to the method, a full-connection neural network architecture is adopted, and boundary condition constraints are introduced into a loss function item for joint optimization training. Through minimization of a loss function, a driving model learns displacement distribution characteristics of an assembly joint surface under thermal deformation, and a mapping relation between a space coordinate point and a corresponding displacement amount is established. The model finally realizes the real-time prediction capability of the thermal deformation field of the assembly joint surface.

The invention relates to a neural network-based adaptive control method for a vehicle formation system. Compared with the prior art, the method solves the defects that a complete target trajectory cannot be known in advance due to the influence of an actual environment and the control direction is unknown due to the unstable state of a vehicle engine. The method comprises the following steps: acquiring tracking data of a vehicle formation system; establishing a dynamic model of the vehicle formation system; establishing an error equation; designing a nusturb type function; designing a radial basis function neural network architecture; designing a GRNN training set and an evaluation index for trajectory reconstruction; designing an event trigger function of an actual controller of the system; carrying out self-adaptive control on the vehicle formation; and controlling the vehicle formation with an unknown target trajectory and an unknown control direction. According to the method, a neural network (NN)-based adaptive control architecture is adopted for a vehicle formation system (VPSs), a real-time trajectory can be predicted online by using a GRNN based on historical data of a target trajectory, and an unknown nonlinear term in the system is compensated, so that the position of a vehicle tracks the predicted target trajectory.

The invention relates to the technical field of network security and data processing, and discloses a violation short message recognition method and system based on deep semantic understanding, and the method comprises the steps: firstly cleaning an original short message, generating a mixed embedding vector through characters, sub-words and pinyin, and carrying out the recognition of the violation short message; then processing through a double-layer detection engine, wherein the first layer utilizes rules and a lightweight model for rapid preliminary screening; in the second layer, for suspected samples, a double-tower fusion neural network architecture is adopted, local and global features are combined, fusion is carried out through a gating unit, and a large language model is input to carry out deep semantic reasoning. The system executes strategies such as interception or flow limiting according to the risk score, and realizes model iteration through a dynamic knowledge base and incremental learning. According to the method, the resource consumption and the detection precision are balanced through the layered architecture, the antagonistic variants are effectively identified by utilizing multi-dimensional feature fusion, and the method has the adaptive evolution capability for a novel violation mode.

The invention provides an aero-engine remaining service life prediction method based on a space-time knowledge graph and SDCNN, and belongs to the field of aero-engine health management. According to the method, a space-time knowledge graph and SDCNN neural network architecture is constructed. According to the method, a spatio-temporal knowledge graph is innovatively constructed for an aero-engine, a BERT model is adopted to carry out data type conversion, and a multi-head graph attention network and a pooling graph attention network complete feature extraction and feature fusion to obtain fusion features; and finally, inputting the fusion features into a stacked expansion convolutional neural network to carry out regression learning on feature data, and then carrying out residual life prediction on the aero-engine. According to the method, modeling and prediction are carried out on complex spatial-temporal characteristic data, the remaining service life of the aero-engine can be effectively predicted under limited data, data support is provided for formulating an aero-engine maintenance strategy, and meanwhile a new thought is provided for predicting the remaining service life of other industrial equipment.

Embodiments described herein provide a Transformer-based neural network architecture that comprises mixture-of-experts time series foundation models to predict different types of time series data. Specifically, given an input multi-variate time series data, a single projection layer may be used to generate patch embeddings for the different time series patterns. The patch embeddings are then passed to a Transformer self-attention layer to compute attention weights, based on which a gating function assigns the patch embeddings into different time series clusters to be further fed to different expert such as feed-forward layers. The feed-forward layers in turn predict a distribution. The output tokens of forecasted time series data are then decoded via the output projection layers from the predicted distribution.

The invention provides a three-dimensional wind field prediction method and system based on multi-modal complementary fusion learning. The method comprises the following steps: S1, acquiring remote sensing observation data and numerical simulation data; s2, obtaining standardized remote sensing features and simulation features; s3, obtaining a unified scene representation; s4, splicing the unified scene representation with the to-be-predicted space-time coordinates, inputting the spliced scene representation and the to-be-predicted space-time coordinates into a physical enhancement decoder, and outputting three-dimensional wind speed vectors at the corresponding space-time coordinates; and S5, iteratively optimizing parameters of the bimodal encoder, the cross-modal attention fusion module and the physical enhancement decoder to form a closed-loop prediction model. According to the method, multi-modal data complementation and physical information deep fusion are realized, through innovating a neural network architecture and a constraint mechanism, the prediction precision under a sparse data condition is remarkably improved, the physical credibility of a result is enhanced, and a technical support is provided for intelligent development of the wind power industry.

A system and method for adaptive neural network architecture implementing sophisticated supervision and signal transmission capabilities. The system comprises a layered neural network monitored by a hierarchical supervisory system that collects operational data and implements architectural modifications. A meta-supervisory system oversees the supervisory process, tracking adaptation patterns and extracting generalizable principles from successful modifications. The system implements novel signal transmission pathways that enable direct communication between non-adjacent network regions through adaptive transformation components and coordinated timing mechanisms. This multi-level approach enables dynamic network adaptation while maintaining operational stability through careful monitoring and controlled modification procedures. The system's innovative architecture allows neural networks to evolve their processing capabilities during operation while preserving reliable performance through sophisticated supervision and controlled signal propagation.

The invention relates to the technical field of digital image identification, and discloses a digital image identification method and system based on artificial intelligence, and the method comprises the steps: constructing a three-layer fusion neural network architecture; analyzing the input image to generate a physical feature vector set; processing an input image to generate a semantic feature vector and a relational graph; on the basis of the physical feature vectors and the semantic feature vectors, physical semantic joint distribution is constructed by utilizing a variational reasoning engine layer; a physical semantic cross attention mechanism is realized; analyzing the deviation degree between the physical semantic joint distribution and pre-established natural image standard distribution; generating an identification result and an interpretability analysis report of the input image based on the deviation degree; according to the method, the normal form transformation from finding forgery traces to verifying naturalness is realized, a brand new theoretical basis and a technical path are provided for the field of digital image identification, and increasingly complicated image forgery challenges can be dealt with.

Apparatuses, systems, and techniques to dynamically identify a neural network architecture for a data point. In at least one embodiment, a data point is fed to a plurality of neural network architectures, in which an optimal neural network architecture is determined based at least in part on information inferenced using the plurality of neural network architectures.

The invention relates to an ORC heat exchanger optimization design method and system based on deep learning. The method comprises the steps that S1, original data samples are expanded based on a data enhancement method; s2, screening key input features of the data samples; s3, synthesizing minority class samples to balance a sample data set; s4, a physical information neural network model is constructed and trained, and heat exchanger performance indexes under different structure parameter combinations are predicted based on the trained model; s5, optimizing the structural parameters of the heat exchanger by adopting a multi-objective coevolution optimization algorithm; and S6, simulation verification is conducted, and the optimal plate heat exchanger design scheme of the target scene is obtained. Through three technical breakthroughs of data enhancement driven by physical constraints, neural network architecture embedded in thermotechnical physics and multi-target collaborative optimization guided by forward distance, systematic technical obstacles in design of the ORC heat exchanger are solved, and an unexpected synergistic effect is generated.

The invention discloses a fault-tolerant neural network optimization method based on automatic architecture search. The method comprises the following steps: introducing a plurality of operators and diversified neural network unit structures in a search space construction process, and constructing a neural network search space supporting multi-dimensional balancing of accuracy, calculation overhead and fault-tolerant capability; in the multi-target architecture optimization process, topological structure coding is performed on a neural network, three performance indexes are introduced in a combined manner based on a Bayesian optimization theory, an agent model is constructed to simulate network performance, and a candidate architecture is generated by using an acquisition function and a multi-target optimization algorithm; in the architecture evaluation and iteration process, candidate architectures are trained through a weight-shared super network, key indexes of the candidate architectures are evaluated, an architecture pool is updated, an optimal compromise is obtained through iteration, and a final neural network architecture is given; the method has high efficiency, flexibility and universality, can be used for automatically exploring the fault-tolerant neural network architecture required by the safety key field, and can adapt to multi-scene and multi-requirement fault-tolerant neural network optimization design.

The invention discloses an underground medium multi-scale forward and reverse modeling method and system based on a decoupling neural network, and belongs to the technical field of underground multi-physics field coupling. The underground medium multi-scale forward and reverse modeling method based on the decoupling neural network comprises the following steps: generating a spatio-temporal evolution data set of multiple physical field variables; a physical field decoupling physical information neural network architecture is constructed, the architecture comprises three special sub-networks, each sub-network takes time-space coordinates as basic input and dynamically receives output of other sub-networks as auxiliary input features, and strong coupling modeling between physical fields is realized through feature sharing and a joint loss function; executing a two-stage training strategy; and executing forward intelligent prediction or key physical property parameter inversion based on the trained neural network architecture. According to the method, a unified neural network framework composed of a plurality of special sub-networks is constructed, physical consistency and data-efficient cross-scale modeling from a rock core scale to a site scale are realized, and forward simulation and parameter inversion are synchronously supported.

The invention discloses a construction site three-dimensional modeling method based on multi-sensor fusion. The method comprises the following steps: synchronously acquiring data by using a laser radar, a camera, an infrared sensor, a GPS (Global Positioning System) and an IMU (Inertial Measurement Unit); the method comprises the following steps of: performing denoising and filtering operation on collected original data, aligning a timestamp and a space of the collected data, fusing the collected multi-source data, and aligning and fusing the processed multi-source point cloud data through a point cloud registration and fusion algorithm to generate a complete three-dimensional point cloud model; generating a grid model by using a surface reconstruction algorithm based on the point cloud data, and performing texture mapping in combination with texture information collected by a camera; the model is simplified and the precision of the model is improved through a model optimization technology, and a reliable three-dimensional visualization foundation for construction management and analysis is formed; real-time detection and modeling of dynamic targets such as mobile equipment and workers on a construction site are realized, and the real-time response capability of the system in a complex construction site environment is ensured through combination of a lightweight neural network architecture and an edge computing technology.

A system and method for adaptive neural network architecture with real-time neurogenesis capabilities during inference operations. The system processes data through a core neural network with integrated supervisory and neurogenesis control systems. A hierarchical supervisory network, comprising low-level, mid-level, and high-level nodes, monitors network activity patterns and information flow. The neurogenesis control system maintains continuous activity maps, detects processing bottlenecks, and determines optimal placement of new neurons using geometric optimization. A modification subsystem implements controlled neurogenesis operations while maintaining network stability. The system handles data through adaptive codeword allocation and fusion of dissimilar data types. This sophisticated approach enables neural networks to dynamically expand their processing capacity during operation, responding to detected bottlenecks while maintaining operational stability through carefully managed integration of new neurons.

The invention belongs to the technical field of nuclear reactor physical numerical calculation, and particularly relates to a method for improving precision and convergence of solving a diffusion equation based on a PINN, and the method comprises the following steps: S1, building a neural network algorithm model combining a physical information neural network PINN and source iteration; s2, aiming at a multi-region smooth transition neutron diffusion problem with small cross section difference between materials, optimizing the neural network algorithm model by adopting an optimization strategy of a training level; and S3, aiming at the multi-region neutron diffusion problem of large neutron flux gradient at the boundary due to large cross section difference between materials, improving the neural network algorithm model by adopting a domain-divided neural network architecture. According to the method, the problem that the calculation cost is exponentially increased in a high-dimensional or complex geometric scene due to the fact that a traditional finite difference method and a finite element method depend on grid division is avoided.

The invention discloses a multimode optical fiber anti-disturbance real-time imaging method based on a double-neural network architecture, which comprises the following steps: firstly, acquiring a speckle image of a target image subjected to multimode optical fiber scattering transmission under a multi-disturbance dynamic state, then extracting spatial domain and frequency domain information of the speckle image, and training through an SFNet classification network to obtain a disturbance dynamic state classification model; and training by using a U-Net imaging network to obtain a speckle imaging model. During testing, the multimode optical fiber disturbance state corresponding to the speckle image is judged by using the classification model, and then the speckle image is input into the speckle imaging model under the corresponding multimode optical fiber disturbance state for reconstruction. According to the method, accurate classification and high-quality reconstruction of the speckle image of the multimode optical fiber under the multi-disturbance condition are achieved, good real-time performance and anti-interference performance are achieved, and the method is suitable for dynamic optical fiber imaging in the complex environment.

The invention discloses a light-weight small target detection method and system for images shot by an unmanned aerial vehicle, and the method specifically comprises the steps: constructing a neural network architecture which comprises a backbone network, a feature aggregation network and a detection head; improvement of light weight and attention enhancement is implemented in the backbone network, and a multi-scale initial feature map is extracted; constructing a feature aggregation network Neck, performing cross-level fusion and refining processing on the multi-scale initial feature map, and outputting a refined feature map; a lightweight target detection head Head is constructed in combination with a large-kernel depth separable convolution module and a special decoupling head structure of a YOLOv11 network, and decoupling prediction is performed on the refined feature map; training is carried out by adopting a mixed loss function based on a normalized Wasserstein distance and modulated IoU, a trained lightweight network is obtained, and detection of a lightweight small target is realized. According to the invention, the complexity of the model is reduced, the detection speed is improved, the high detection precision is maintained, and the method is suitable for real-time detection tasks on an unmanned aerial vehicle resource limited platform.

The invention relates to the technical field of PCB (Printed Circuit Board) blind hole defect detection, and discloses an X-ray-based PCB blind hole defect detection method, which comprises the following steps of: irradiating a PCB by using X rays, performing data acquisition by using CT (Computed Tomography) projection, establishing an initial image, reconstructing the initial image by using filtered back projection to obtain a reconstructed image, and detecting the blind hole defect of the PCB according to the reconstructed image. According to the blind hole defect detection method based on the deep residual network, through the double-flow neural network architecture constructed on the basis of the deep residual network, geometric features and material features are comprehensively utilized, the blind hole defect detection accuracy is improved, the blind hole defect detection accuracy is improved, and the blind hole defect detection accuracy is improved. The PCB blind hole defect detection device can comprehensively and accurately detect and classify various defects of PCB blind holes, such as cracks, wrinkles, recesses and filling holes, improves the efficiency and accuracy of defect detection, is helpful for timely finding quality problems in the PCB production process, and guarantees the product quality.

The application belongs to the technical field of medical signal processing, and discloses a sleep apnea detection method and system, an electronic device and a storage medium. The method comprises the following steps: acquiring a synchronous ECG signal and a breathing signal, and processing the synchronous ECG signal and the breathing signal into a bimodal data segment containing multiple time scales; constructing a parallel neural network architecture, extracting ECG features and breathing signal features from each scale data segment respectively, and forming a cross-modal feature pair; designing a cross-scale dynamic weight correction attention mechanism, introducing a correction factor based on the Euclidean distance between feature vectors, dynamically weighting and information fusion on the cross-modal feature pair between different scales, so as to enhance the relevance of local details and global context; inputting the fused multi-scale features into a classifier, and outputting the probability of a sleep apnea event. Through multi-scale collaborative analysis and a dynamic weight correction attention mechanism, the application effectively fuses physiological information of different time dimensions, and improves the detection accuracy.

A computer system for adaptive neural network architecture implementing sophisticated supervision, pruning, and signal transmission capabilities. The system operates a layered neural network monitored by a hierarchical supervisory system that collects activation data, identifies operation patterns, implements architectural changes, detects network sparsity, coordinates pruning decisions, and manages resource redistribution. A meta-supervisory system tracks supervisory behavior patterns, stores successful modification and pruning patterns, and extracts generalizable principles from these patterns. The system manages signal transmission pathways that enable direct communication between non-adjacent network regions through signal modification and temporal coordination. This multi-level approach enables dynamic network adaptation and efficient resource utilization through pruning while maintaining operational stability. The system's innovative architecture allows neural networks to evolve their processing capabilities during operation while preserving reliable performance through sophisticated supervision and controlled modification.

The invention discloses a rainfall image sequence similarity search method based on a twin neural network. The rainfall image sequence similarity search method comprises the steps of obtaining a target rainfall image sequence and a target rainfall image library; constructing a rainfall image similarity model; based on the target rainfall image sequence and the target rainfall image library, performing training optimization on the constructed rainfall image similarity model to obtain a target rainfall image similarity model; and based on the target rainfall image sequence and the feature vector of the corresponding frame in the target rainfall image sequence extracted by the target rainfall image similarity model, carrying out rainfall image sequence similarity search on the target rainfall image library to obtain a rainfall image sequence similar to the target rainfall image sequence. According to the method, deep learning and a meteorological mechanism are fused, efficient similarity search of a rainfall image sequence is realized by constructing a rainfall image sequence similarity search model of a feature extraction network and twin neural network architecture, and three bottlenecks of feature robustness, time sequence modeling and calculation efficiency are broken through.

The invention aims to provide a high-precision design and optimization method for an EMI (Electro-Magnetic Interference) filter aiming at the corresponding defects in the prior art, a mixed neural network architecture model is constructed by combining a feedforward neural network and a convolutional neural network, and the filter is subjected to reverse modeling by utilizing the mixed neural network architecture model, so that the filtering precision is improved. According to the method, the predicted value, the topological structure and the element parameters of the filter are obtained, then the element parameters are subjected to multi-objective optimization by using a multi-objective genetic algorithm to obtain the optimized element parameters, and the optimized element parameters and the topological structure of the filter jointly form an optimization scheme of the filter, so that the design efficiency and the optimization precision of the EMI filter can be remarkably improved; and the design period of the EMI filter is shortened, and the market demand of rapid iterative updating of current high-frequency electronic equipment can be met.

The invention relates to the technical field of data processing, and discloses a low-altitude radar information processing system based on big data analysis. The system comprises a data fusion module, a false alarm suppression module, a measurement target identification module and a grading early warning mechanism establishment module. Firstly, low-altitude radar signals are collected, space-time reference synchronization is carried out, original low-altitude radar data are obtained, and then data association matching is carried out by using a clustering fusion algorithm; secondly, optimizing the threshold value by using a multi-strategy fusion particle swarm algorithm, and performing false alarm suppression; establishing a three-path feature network model based on a multi-modal neural network architecture, carrying out target recognition, and outputting a low-altitude radar measurement target recognition result; and finally, according to a low-altitude radar measurement target identification result, carrying out environment constraint integration, generating a radar target motion track, and establishing a grading early warning mechanism. According to the method, low-altitude radar data are processed and modeled, the purpose of low-altitude radar information processing is achieved, and the method is accurate and objective.

The invention provides a converter valve key component burning defect identification method based on a neural network, and belongs to the technical field of power electronic equipment fault diagnosis, and the method comprises the steps: collecting multi-modal data through an infrared thermal imager and a plurality of sensors, and inputting the pre-processed multi-modal data into a specially designed neural network architecture; the framework comprises an image feature extraction module, a time sequence feature extraction module, a feature fusion module and a classification positioning module. According to image processing, improved ResNet50 is combined with an attention mechanism, time sequence features are extracted through a bidirectional long-short-term memory network and a time convolution network, and effective feature combination is achieved through a dynamic weight fusion mechanism. Meanwhile, a double-phase heat conduction model is introduced to analyze temperature distribution, accurate identification of burning defects of key components of the converter valve in a complex environment is realized through large-scale data set training and a two-stage optimization strategy, and key technical support is provided for safe operation of a power system.

The invention discloses a small sample crack recognition method and system based on transfer learning, and the method comprises the steps: constructing an innovative and deep-coupled neural network architecture, and enabling a self-adaptive crack perception attention module integrated with a parallel asymmetric convolution kernel to recognize the linear geometric features of a crack, and generating an attention graph; then, the attention map is adopted to carry out pixel-by-pixel signal pre-modulation on an original input image, the enhanced image is sent to a pre-training MobileNetV4 backbone network integrated with a feature-level linear modulation layer, and channel-level dynamic adaptation of feature flow in the network is realized by learning affine transformation parameters; and finally, the extracted depth features are sent to a Gaussian naive Bayes classifier for classification, and depth geometric analysis is carried out on the attention map so as to realize interpretable fracture severity evaluation. According to the method, the accuracy, robustness and interpretability of crack identification are remarkably improved, and the technical bottleneck in a small sample scene is effectively solved.

The invention provides a product full life cycle management method based on digital twinning, a medium and a product full life cycle management system based on digital twinning, and belongs to the technical field of digital twinning. Based on a super-sparse pre-training model of a liquid neural network architecture, in combination with spectral clustering and an attention mechanism, precise prediction of product performance degradation is realized, and intelligent maintenance suggestion generation and dynamic optimization of operation parameters are realized through a Hilbert matrix state evaluation and decision adjustment mechanism. The balance between calculation efficiency and prediction precision is realized by using a multi-precision simulation analysis and incremental updating technology, a backtracking decision matrix is established to support the optimal decision of product decommissioning and recovery, and the technical problem that the real-time perception and dynamic prediction optimization management of the full-life-cycle multi-dimensional state of the product cannot be realized in the prior art is solved.

The invention belongs to the technical field of data processing, and particularly discloses a multi-source heterogeneous data acquisition access method and system based on intelligent mapping and a dynamic interface, and the method comprises the steps: registering an input data source class, generating a data source identifier corresponding to the data type, and storing the data source identifier; respectively executing meta-model analysis operation to generate a standardized meta-model associated with the data source identifier and a container instance template; generating a mapping rule set through a twin neural network architecture with multi-modal feature fusion and a preset mapping rule; a middle layer API interface is constructed through interface logic, and real-time conversion and access between input and output data are achieved; and realizing containerized parallel access based on the container instance template and the containerized interface service instance. According to the method, data source change is dynamically adapted, the real-time performance and accuracy of the mapping rule are ensured, and meanwhile, multi-source heterogeneous data can be accessed efficiently in real time in combination with containerized parallel processing and a data source change monitoring mechanism.